My personal (hopefully, everyone else's) perspective of digital libraries is that digital libraries will not replace traditional libraries, the two kinds of libraries will complement each other, and digital libraries will provide new possiblities and paradigms for knowledge dissemination, collaborative learning, and social services. Under these assumptions, linking user-learner needs and behavior to digital library design should be the foremost requirement in the development of digital libraries. So far, user interface/usability and information retrieval system design/ implementation have not been efficiently linked. As I see it, there are at least three issues. First, all possible users and learners should be included in the overall planning and design because digital libraries have significant social impacts. Some authors have already voiced their concerns about the ''cyberia'' (the have's and have-not's) issue of the Internet. Second, in order to avoid any unforeseeable social hazardous impact or technical breakdown, the development of digital libraries should not be driven by technologies alone. We have seen the impacts of autos, freeways, chemicals, and many more. Third, user-learner needs should be integrated into the overall planning and design, leading to a system-level evaluation and implementation. The following are the associated challenges and opportunities:
Challenges: (1) developing an iterative system-level design and user-learner evaluation process of digital libraries, (2) jointly modeling user-learner behavior and digital library system behavior, (3) designing programmable and reusable digital libraries.
Opportunities: (1) developing integrated digital library research on user-learner groups, multi-culture/social context, and application domains, (2) building reusable software tools for digital libraries and other NII applications, (3) advancing research on human-centered enabling technologies for digital libraries.
Let me now discuss the challenges and opportunities in a little more detail.
Software engineering projects are planned, designed, implemented, and managed through an iterative process by a team of users, clients, system managers, designers, and engineers. Digital libraries will be involved with complex, large configurations of hardware, software, middleware, and infoware. Iterative design and evaluation of digital libraries will be essential in this case.
Although user-learner behavior and information retrieval system behavior have been studied, they need to be jointly modeled. Before a large system is implemented, some field data collection, model testing, and theory building will be necessary. Through electronic data logging, we have the capability to collect real-world data beyond traditional user evaluation methods, such as interview, video, and direct observation.
Due to dynamic user-learner behavior and long-range social effect, key components of digital libraries should be programmable and reusable. We should take into serious consideration the life span of library content and the interoperability of library management systems. In computer utilization, the legacy issue of hardware and software has been significant. We should also address this issue for infoware (information content).
In various research areas, such as library science, computer science, information science, social science, humanity, history, and arts, we have not had the chance to develop interdisciplinary research at the scale of digital libraries. Digital library research offers real-world testbeds for conducting and evaluating such research activities. Supplementary funding for such research activities should be suggested to federal agencies.
Software tools developed for digital libraries will be useful to other NII applications. Current object-oriented technologies for enterprise systems, such as CORBA and Compound Document, will be very good starting points for the development of digital libraries. Sharing software tools for research should be encouraged and supported.
Digital library design suggests a wide variety of needs of human-centered enabling technologies. An important issue is how to transform traditional computer science and information technology research to research with additional HCI (human computer interaction) emphasis?